Digital video recorders are commonly used to record digital video for transfer to a computer. Once on the computer, users may edit, enhance, and share the digital video. However, today's digital video recorders compress digital video using forms of encoding that use temporal compression. That is, the compressed video includes predictive (P) and bidirectional (B) frames that are not actual images, and instead are only mathematical data representing the difference between an index (I) frame that is encoded as an image.
Temporal compression enables compression of digital video to smaller file sizes on the camera, but creates a multitude of problems for users that want to transfer the video to their computers in order to work with the video. Because the P and B frames are only defined by reference to other frames, they must be transcoded in order for a user to edit them. This transcoding generally takes place upon import of the digital video from the camera.
FIG. 1 illustrates a prior art system with a video camera 105 and a computer 110. The video camera 105 captures and stores a video file 115 having a size X. This video is encoded using temporal compression. Upon transfer from camera 105 to computer 110, the video must be transcoded (to remove the temporal compression) and stored. The resulting file 120 has a size of 3X to 10X, and thus is much larger than the original file on the camera. Because of these expansions, it does not take that much video for the size of the file to become prohibitive for most users. Furthermore, the transcoding is a time- and computation-intensive process. Transferring 30 minutes of video can take 90 minutes due to the transcoding. Accordingly, there exists a need for a video camera with the capability to record video that is not temporally compressed without sacrificing quality or creating excessively large file sizes.